Join 3,438 readers in helping fund MetaFilter (Hide)


Aids in Africa
December 12, 2003 6:23 AM   Subscribe

Aids in Africa - you know the facts right? Well perhaps not, what you know are the predictions of a Computer Model. Rian Malan in today's Spectator highlights how alarmingly inaccurate such models are proving. Paul Henman illustrates how common it is to build political assumptions into a model and then hide them under layers of complexity and apparent objectivity. Think global warming. How do we challenge the models that increasingly determine our opinions and priorities?
posted by grahamwell (15 comments total)

 
It's an interesting article. I don't really know enough about statistical analysis to comment on it's veracity...but I do think he raises an interesting point about the discussion of AIDS.

In a similar way that criticizing Israel is often perceived as anti-Semitic, people who criticize AIDS projections and data models are often perceived as heartless or intentionally cruel.
posted by dejah420 at 7:46 AM on December 12, 2003


Here's a question: AIDS is the acronym but I see 'Aids' all the time. Is it an acceptable variant of the name?
posted by Dagobert at 7:54 AM on December 12, 2003


Certainly something to think about, upon first perusal. Thanks, grahamwell.
posted by PigAlien at 7:56 AM on December 12, 2003


Think global warming.

Ah, a psychoceramic simile.
posted by y2karl at 9:23 AM on December 12, 2003


Michael Crichton makes a similar argument against using computer models to determine international policy.
posted by event at 11:02 AM on December 12, 2003


I keep hearing about the AIDS Crisis in Botswana and wondering, because I have family there who describe it as being about as good as it gets in SubSaharan Africa. In fact, they may be coming back this year because my uncle may be losing his job to the mandatory retirement age of 60(!), which speaks to me of overpopulation, not under.
posted by hob at 11:35 AM on December 12, 2003


If you want to talk Global Warming, I have done some programming on global climate models used to predict this stuff, and they are very primitive. Just the parts I was working on, relating to cloud formation, state of the art is considered 10km grid squares. A cloud is usually a lot smaller than 10km. This is only one tiny example of the deficiencies of computer models in predicting climate. There are just so many feedbacks to the system that aren't modelled that it's impossible to have any confidence in their predictions. Realistically what we know is that over the past fifty years, the earth has been warming, and simultaneously, CO2 levels have been rising. We don't know what will happen if CO2 levels stay the same or keep rising.
posted by cameldrv at 1:32 PM on December 12, 2003


Ok, I'll say it - I think the whole article is deplorable. There, see? Was that so hard??

It's irritating to read about a man who thinks the whole thing is just a numbers game, as to say that the death rate isn't nearly high as it's modeled out to be. What's the point in that? Is he saying we shouldn't bother with it? Is he saying it's unfortunate but it's really not that big of a deal? What's his point - that the numbers are too high?? So what??!!! I say pick your cause and run with it. Maybe he could try becoming a champion in the fight against malaria intead of being the naysayer of AIDS.

It's annoying when someone offers up nothing but problems, and then doens't make any effort to offer a fix.
posted by matty at 2:11 PM on December 12, 2003


Matty comments:

"It's annoying when someone offers up nothing but problems, and then doens't make any effort to offer a fix."

Why? Why is the person who proclaims that the emperor has no clothes required to then find him attire?

I'm personally very big on finding a cure and/or long-term treatment for AIDS/HIV because I believe that the benefits are great, not just in treating the disease but for medical research in a general sense (i.e., the "spinoff" effect of increasing knowledge as an aside to the stated aim of research). However, I don't see the value in uncritically swallowing data on the disease that may be incorrect.

Personally, I like having my data on anything resemble truth as much as possible -- for the reason that truth and accuracy is, in itself, a laudable goal.

Knowing more accurate numbers (if indeed these are) doesn't make me any less interested in finding a cure. It does make me mildly relieved that not nearly as many people are dead of the disease as suspected. It's not a pleasant way to die.
posted by jscalzi at 3:04 PM on December 12, 2003


A long time ago, I was involved in a research study of the actual effect of factors that are believed to decrease population: war, historical famine, and plague disease

Historical famine was the strongest population reducer, as it could take an area over 100 years to recover economically enough to support increased population. ("Historical" to distinguish it from man-made modern famines, or artificial population increases based on foreign aid.)

War was ruinous, but often resulted in a population boom afterwards, with declining prosperity.

Plague disease was the most interesting, because not only did the population recover quickly, and with some modest increase (as the columnist mentioned), but it also resulted in increased prosperity!

The reason being that wealth is concentrated with the survivors. So individual farmers have larger farms, with extra crops for sale; young people get their parents money earlier, which they put to good use; prices go up, but wages go up faster; and there is greater political stability.

Two cases in point--the second great European black plague was followed by the Renaissance, and the third by the Industrial Revolution.

It is questionable to apply this model to the AIDS epidemic, other than as a possibility, but if AIDS is as widespread as suggested, it could result in a better life for the survivors.
posted by kablam at 3:50 PM on December 12, 2003


Here's a question: AIDS is the acronym but I see 'Aids' all the time. Is it an acceptable variant of the name? -- Dagobert

British English distinguishes (not always consistently) between acronyms which are pronounced as letters (which are almost always capitalized), and acronyms which are pronounced as words (which are usually, but not always, shown in mixed case). The Guardian has an online style guide showing Aids; the Economist doesn't mention Aids but shows several examples.

posted by dhartung at 10:40 PM on December 12, 2003


The person who points out that the emporer has no clothes should be politle enough to offer him some new ones right after he's done embarrassing him.

And that artice was anything BUT uncritical - it went out of it's way to point fingers and find flaw.
posted by matty at 12:26 AM on December 13, 2003


Since some are so concerned about the emporer's coture...

BURKE
"It was a bad call, that's all."

Ripley snaps. She slams him against the wall, surprising
herself and him, her hands gripping his collar.

RIPLEY
"Bad call? These people are fu##*ng
dead, Burke!"

It's not just numbers and data, people. And it doesn't matter if the numbers are 'higher' than they might really be. That kind of apathy and convenient removal of one's self from the problem will kill you in the end every time.
posted by matty at 12:38 AM on December 13, 2003


having also done a bit of modeling i'll agree that the underlying assumptions can become controlling factors in the final outcome. However, the best way to solve this is through validation and then changing your original assumptions. the problem is that as you get more factors effecting the overall prediction the less likely you are to have a representative fit or get a solution at all (in english (i hope): if i only have one or two variables it's easy to fit an equation (though badly), if i have 10 or 20 they all interact and you don't know if your getting garbage for values)).

so how do we do validation in this case. In order to validate hydrological models (water flowing from where it lands to where it goes through a river) you need to take data...lots and lots of data. Even with the (limited) amount of data we have to work with we still have a hard time modeling it. however, the idea still holds, you take as much data as possible and then you try to use it to the best of your abilities so that you can better fit the model and make more likely predictions (weather just ends up being a really hard thing to predict). So basically we need to do some head counts people.

Pragmatically we're not going to be worse off putting more money into aids research or limiting pollution of the natural environment, so no harm no foul. We all love malthusian processes until we're involved in them.
posted by NGnerd at 7:44 PM on December 13, 2003


NGnerd: Here's the $64 question: what do you do with perhaps 50 million *excess* males? Between India and China, there are projected to be that many men, or more, who will not have wives, or jobs, or much prospects.
In other words, now that humans outnumber rats, we are the #1 target for mother nature to come along with some "population reducer", to put us back into *whatever* is the natural balance.

And, as I mentioned before, plague seems to be the most painless way of reducing our numbers.
posted by kablam at 8:18 PM on December 13, 2003


« Older Make sure to use lots of lube...  |  29 May 1453, Constantinople fe... Newer »


This thread has been archived and is closed to new comments